To use an ATM you must first locate an audio jack on the ATM machine and must insert a standard 3.5 mm headset into the audio jack to enable talking mode. A welcome audio message will be heard. For this step the visually impaired consumer has to first locate the headphone jack itself, which is a cumbersome and time consuming procedure in itself.

The ATM keypad is identical to a standard telephone keypad with raised dot on number 5. All audio instructions are in the form of Interactive Voice Response (IVR) which will guide you through the process of ATM transaction. Towards the right side of the ATM keypad you will find a column of important keys which are:

Cancel: with the universal tactile shape ‘X’. Enter: with the universal tactile shape ‘O’. Clear: with the tactile shape ‘I’ or ‘<’.

These are a pre programmed set of instructions/responses which lacks the dynamic conversations and needs of the consumer.

Not Secure, if the impaired person takes assistance from someone, the helper may trick him and withdraw money more than what is requested by the impaired person since there is no audio notification.

For the uneducated consumers the current ATM solution supports two language English and Hindi majorly, and it assumes that the consumer is fluent in reading these languages, which can be challenging for the uneducated masses. We propose a solution which can be transformed to support regional languages and at the same time only requires the consumer to be able to talk in any language of his or her choice.

We propose a Machine Learning based solution, which will be leveraging technologies like Facial Recognition and Natural Language Processing. The following are the steps for our solution -

1. The banks during the issuance of ATM Card can collect special information for the uneducated/visually impaired, this will comprise of Biometrics (fingerprints) and Face ID in the form of Photographs of the consumer. The consumers can select the languages of their choice at the time of registration and the bank will also provide them with the ATM pin for their ATM Cards.

2. When the user inserts the card into the ATM, the database can check if this card corresponds to a visually impaired/uneducated customer, after this step the languages selected by the consumer will be used to communicate with him.

3. For the visually impaired consumer, we will further run a Facial Recognition Check (which will compare the current user with the Face ID collected at the time of registration). If the check confirms we will proceed with the Biometric Validation to ensure absolute security (fingerprints are easier for the visually impaired consumers as they don’t have to rely on typing anymore).

If in any case another user tries to use a card corresponding to a visually impaired consumer (which can be checked as the current user will fail both the Facial Recognition and the Biometric Validation), he/she will be required to enter the ATM pin for using the account corresponding to that card. Like in the case of a family member of the visually impaired customer trying to use the ATM Card, hence the system is still capable of functioning as a conventional ATM system currently in use.

This step is same for the Uneducated Consumers, they will also undergo the Facial Recognition and Biometric Checks to access the ATM.

4. Now the ATM will enable the Dialogue Flow specific to the needs and requirements of the consumers (by using the languages selected at the time of registration), the consumers can tell the machine to withdraw a certain amount and select the account (Savings, Current Bank Account) over voice commands. The consumer will be provided with an experience which is similar to a normal human conversation and that too in the language of their choice.

At the end of the transaction the ATM will generate both audio and video signals in the form of messages / flash alerts to inform the user that his session at the ATM is complete and he is securely logged out of the ATM, users can acknowledge the same by simply chatting with the machine.

The process will empower the consumers who have special needs and require careful assistance, by making them self reliant and simplifying the process of ATM Transactions.

Anudish Jain 2016UEC2120